From 4e88225c67d776eac2ee779bf3fbc1c54db080d3 Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Wed, 26 Dec 2018 17:38:06 +0100
Subject: [PATCH] support indexing by client

---
 app/models/doi.rb  | 11 ++++++++---
 lib/tasks/doi.rake |  3 ++-
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/app/models/doi.rb b/app/models/doi.rb
index 8d3957124..6ec3b0b20 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -443,10 +443,11 @@ def self.index(options={})
     from_date = options[:from_date].present? ? Date.parse(options[:from_date]) : Date.current
     until_date = options[:until_date].present? ? Date.parse(options[:until_date]) : Date.current
     index_time = options[:index_time].presence || Time.zone.now.utc.iso8601
-
+    client_id = options[:client_id]
+    
     # get every day between from_date and until_date
     (from_date..until_date).each do |d|
-      DoiIndexByDayJob.perform_later(from_date: d.strftime("%F"), index_time: index_time)
+      DoiIndexByDayJob.perform_later(from_date: d.strftime("%F"), index_time: index_time, client_id: client_id)
       puts "Queued indexing for DOIs created on #{d.strftime("%F")}."
     end
   end
@@ -455,13 +456,17 @@ def self.index_by_day(options={})
     return nil unless options[:from_date].present?
     from_date = Date.parse(options[:from_date])
     index_time = options[:index_time].presence || Time.zone.now.utc.iso8601
+    client_id = options[:client_id]
 
     errors = 0
     count = 0
 
     logger = Logger.new(STDOUT)
 
-    Doi.where(created: from_date.midnight..from_date.end_of_day).where("indexed < ?", index_time).find_in_batches(batch_size: 250) do |dois|
+    collection = Doi.where(created: from_date.midnight..from_date.end_of_day).where("indexed < ?", index_time)
+    collection = collection.where(client_id: client_id) if client_id.present?
+
+    collection.find_in_batches(batch_size: 250) do |dois|
       response = Doi.__elasticsearch__.client.bulk \
         index:   Doi.index_name,
         type:    Doi.document_type,
diff --git a/lib/tasks/doi.rake b/lib/tasks/doi.rake
index c97a8e520..fb8ee4ac0 100644
--- a/lib/tasks/doi.rake
+++ b/lib/tasks/doi.rake
@@ -61,8 +61,9 @@ namespace :doi do
     end
 
     index_time = ENV['INDEX_TIME'] || Time.zone.now.utc.iso8601
+    client_id = ENV['CLIENT_ID']
 
-    Doi.index(from_date: from_date, until_date: until_date, index_time: index_time)
+    Doi.index(from_date: from_date, until_date: until_date, index_time: index_time, client_id: client_id)
   end
 
   desc 'Index DOIs per day'